NASA Technical Memorandum !_02418 .... -.....

' Thermal Modelling of Various_:___._:-_ ;_=Therma! Barrier Coatings in a

--_:_-_- _--High Heat Flux Rocket Engine

James A. iN6sbitt

Lewis Research Center

Cleveland, Ohio

........ December 1989

IW A..... (NASA-TM-!0241_) THERMAL MODELLING OF

VARIOUS THERMAL _ARRIER COATINGS IN A HIGH

HEAT FLUX ROCKET ENGINE (NASA) 31 pCSCL i

Nq0-15911

IF Unclas

G3/26 O254_51

https://ntrs.nasa.gov/search.jsp?R=19900007595 2020-03-20T00:15:27+00:00Z

O0

L_

iW

THERMAL MODELLING OF VARIOUS THERMAL BARRIER COATINGS IN

A HIGH HEAT FLUX ROCKET ENGINE

James A. Nesbitt

National Aeronautics and Space AdministrationLewls Research Center

Cleveland, Ohio 44135

: _-_- L J _ SUMMARY f- i_,- F'c___,c_-i .... _ J

Tradltlonal. _PS_ ZrO2-Y203iTBC's_and APS and I_PPS ZrO2-Y20_/NI-Cr-AI-Y

cermet coatings _ere teste_ In a H2/O 2 rocket e_'Ine._oeated at the NASA Lew_sResearch Center_ The traditional ZrO2-Y2Q 3 thermal 6arrler coatings (TBC s)_showed considerable metal temperatur_ redOctlons during testing In the

hydrogen-rich environment. A thermal model was developed to predict the

thermal response of the tubes with the various coatlngs. Good agreement wasobserved between predicted temperatures and measured temperatures at the Inner

wall of the tube and In the metal near the coating/metal Interface.

The thermal model was also used to examine the effect of the differences

In the reported values of the thermal (:onductlvlty of plasma sprayed ZrO2-Y203ceramic coatings, the effect of a lO0_m (0.004 In.) thick metallic bond coat,

the effect of tangential heat transfer around the tube, and the effect of radi-

ation from the surface of the ceramic coating. It was shown that for the

short duration testing in the rocket engine, the most Important of these con-

siderations was the effect of the uncertainty In the thermal conductivity of

traditional APS ZrO2-Y203 TBC's which can amount to significantly different

temperatures (_,lO0 °C-) predlcted In the tube. The thermal model was also usedto predict the_thermal response of a coated rod in order to quantify the dlf-

ference In the_metal temperatures between the two substrate geometries and to

explain the prevlously-observed Increased llfe of coatings on rods over that ontubes. ,-,_ '_'_. _, l___,,,

A thermal model was also developed to predict heat transfer to the lead-

ing edge o_,_PFT_blades during start-up of the space shuttle main engines.

The abilit# of various TBC's to reduce metal temperatures durlng the two ther-

mal excurs_Qns occurring on start-up was predicted. Temperature reductions of

150 to 470'_-C were predicted for 165 wm (0.0065 in.) coatings for the greaterof the two thermal excursions _"

INTRODUCTION

The blades In the hlgh-pressure fuel turbopump (HPFTP) on the space shut-tle maln engines (SSME) undergo extreme thermal translents on start-up of theengines. The thermal shock to these blades is believed to contribute to crack

formation and/or crack propagation which has led to the early replacement of

the blades far short of the desired design life. During early development

work on the SSME, thermal barrier coatings (TBC's) were examined as a means of

reducing the thermal shock to these blades. However, spallatlon of the ceramic

top layer during test flrlngs resulted In dropping from consideration the use

of ceramic coatlngs. Currently, only a metallic coating, providing mlnlmalthermal protection, is applied to HPFTP blades.

Recent advances In TBC technology prompted a re-examlnatlon of the use ofthermal barrier coatings for HPFTP blades (refs. 1 and 2). A hydrogen/oxygenrocket engine was developed at NASA Lewis Research Center to test the dura-bility of various TBC's in a high heat flux environment. Numerous thermalbarrier coatings on tube and rod substrates were tested (ref. 2). This studyfound that, In general, coatlngs were more durable on rod substrates than ontubes. This increased durability can be attributed to lower coating and sub-strate temperatures due to the greater thermal mass of the rod, although tem-perature measurements on rod substrates were not performed to quantify thistemperature reduction.

The purpose of the present work was to examine the thermal transport andevaluate the thermal protection of several different TBC's tested in the

rocket engine at NASA Lewis. These coatings included traditional TBC's with a

ZrO2-Y203 top coat and Ni-Cr-AI-Y bond coat, as well as two cermet coatings

consisting of different amounts of ZrO2-Y203 and NI-Cr-AI-Y in the outer layer

above a Ni-Cr-AI-Y bond coat. One cermet coating was applied by low pressure

plasma spraying (LPPS) while the second was applied by air plasma spraying(APS). Metal temperatures were measured either below the coating or on the

inner wall of the tube. A thermal model was developed to predict temperatures

In the coating and substrate for both tube and rod geometries. Use was made of

previously measured gas temperatures and heat transfer coefficients for tubes

located at the throat of the rocket engine (ref. 3). Using this data, prevl-

ous modelling work showed good agreement between measured and predicted near-surface metal temperatures in an uncoated tube (ref. 3).

The thermal model was also used to examine the effect of the range In the

reported values of the thermal conductivity of plasma sprayed ZrO2-Y203 ceramiccoatings, the effect of a lO0 pm (0.004 in.) thick metallic bond coat, theeffect of tangentlal heat transfer around the tube, and the effect of radia-

tion from the surface of the ceramic coating. Comparing predicted temperature

profiles for rod and tube substrates allowed the temperature differences

between the two substrates to be quantified.

The effectlveness of the various coatings In reduclng the thermal shock to

SSME-HPFTP blades was examined by predicting surface temperatures at the lead-

Ing edge wlth and without the coatings. This latter study was accomplished by

developing a thermal model for the leading edge and using gas temperatures and

heat transfer coefficients appropriate for SSME start-up.

EXPERIMENTAL PROCEDURE AND RESULTS

Rocket Engine Operation

The rocket engine is shown schematically In flgure I. Five 0.953 cm(0.375 in.) tubes (or rods) were held in a copper specimen holder such that allgas flow was perpendicular to the cylindrlcal axls. The hot gas exiting thecombustion chamber is constricted by the presence of the tubes located at thethroat of the rocket engine. High pressures in the combustion chamber resultin Mach ] exhaust gas velocities slightly upstream from the throat plane (theplane passing through the cylindrical axis of each of the tubes yielding thesmallest cross-sectional throat area). The area blockage ratio for the engine,given as W/(W - D) (see fig. l), is equal to 2.875 giving an area reduction(Inverse of the area blockage ratlo) at the throat of 0.348.

Only three of the flve specimens are fully exposed In the gas flow (posl-tlons 2, 3, and 4, fig. I). However, gas temperature measurementsIndlcatedsignificant temperature differences between these three center test poslt|ons.Hence, all testing of thermal barrler coated tubes and rods was performed wlthtest specimens in position 2. Along the length of the tube at position 2, thetemperature changed significantly, generally being highest at the mldpolnt anddecreasing toward the rocket chamber walls. Consequently, metal temperaturesin the tubes were only measured at the midpoint (lengthwise) of the tube. How-ever, even at the mldpoint, the gas temperature varied with angular positlonaround the test tube such that gas temperatures measured near the stagnatlonpoint of the tube (e = 0 °, facing the injector) were slgnlf_cantly greater thangas temperatures measured on the exhaust side of the tube, opposite the injec-tor (e = 180°). As a consequence, Instrumented tubes were always oriented suchthat the thermocouples faced the injector (e : 0°). Typical gas temperaturesfor the stagnation polnt and on the exhaust side of the tube are shown in flg-ure 2. These relatively low gas temperatures were achleved by malnta_nlng ahydrogen-rlch environment in the rocket chamber with an oxygen to hydrogen massratlo of approximately 1.2 to 1.4 (stolchiometric combustion at a mass rationear eight produces gas temperatures in excess of 3000 °C). Gas temperaturemeasurements are discussed in detail elsewhere (ref. 3).

Standard operation of the engine Involved opening the 02 valve 0.1 sec

before opening the H2 valve wh|ch generally ensured engine ignition In thehydrogen-rich environment. The test duration was 1.2 to 1.3 sec. At the end

of the test, the 02 valve was closed 0.3 sec prior to the closing of the H2

valve which resulted in a short H2 purge of the test tube. As either the 02 or

the H2 valve was closed, each llne was purged by gaseous N2 which resulted In

a N2 purge coollng the specimen to room temperature. The pressure In the cham-ber of the rocket engine was malntalned at approxlmately 2.07 MPa (300 psi).

Speclmen temperatures before, during and after a test were measured at a fre-

quency of lO0/sec. Further details of the rocket engine operatlon are givenIn reference 3.

Thermal Barrier Coatings

Tradltlonal TBC's applied at NASA Lewis. - Near-surface metal temperatures

were measured at an angular orlentatlon facing the injector on 0.953 cm o.d.,

0.648 cm l.d. (0.375 in. o.d., 0.255 In. I.d.) directlonally solldifled (DS)Mar-M246+Hf tubes I coated wlth a tradltlonal plasma-sprayed ZrO2-Y203 TBC. The

tubes were Instrumented by cutting channels 0.038 cm (0.015 in.) in depth by0.056 cm (0.022 In.) In width. IndlvlduaIIy-sheathed thermocouples, 0.0254 cm

(O.OlO In.) diameter sheath with 0.0065 cm (0.003 in.) diameter wire, were

placed in the channels and the beads welded to the bottom of the channel. Thechannels were covered with a 0.013 cm (0.005 In.) Incone] 600 sheet which was

welded In place to seal the channel from the hot gas. A cross-sectional view

of the channel and the Individually-sheathed thermocouple wires Is shown in

figure 3. Prlor to coating application, the surface of the instrumented tubes

was cleaned and roughened by grit blasting with 60 grit, hlgh-purlty alumina at0.48 MN/m 2 (70 psi) followed by ultrasonically degreaslng. The instrumented

tubes were then coated wlth a 25 to 50 _m (O.OOl to 0.002 In.) APS NI-Cr-AI-Y

IThe compositlon of DS Mar-M246+Hf is Ni-10Co-9Cr-lOW-5.5AI-2.5Mo-l.5Ta-1.5Ti-O.15C-O.O15B-O.O5Zr-1.75Hf wt %.

bond coat layer. Bond coat roughness before top coat application was12.2 pm RA (481Nin.). The ceramic top coat was plasma sprayed to thicknessesranging from I00 to 275 _m (0.0045 to 0.0095 in.). Details of the plasmaspraying process and initial powder composltions are given In table I and spec-imen designations, coating thicknesses, and surface roughnesses are given intable II.

Two of the coated tubes were tested in the as-sprayed condition, afterwhich the surface was smoothed by sanding with SiC paper and the coatings wereretested. The surface roughness of the as-sprayed coatings and that followingsanding is given in table II. It was previously shown that differences in thesurface roughness in the ranges shown in table II for as-sprayed and sandedsurfaces have little effect on the heat transfer to the tubes (ref. 3). A

traditional ZrO2-Y203 TBC was also applied in a manner as that given above buton a B-1900 superalloy substrate. 2 A single thermocouple was attached to theinner wall of this tube to permit temperature measurements at the same locationas for the cermet coatings, discussed below. The coating designation (TK3) andthickness for this specimen are glven In table II.

Post-test microstructures of the traditional TBC's plasma-sprayed at NASA

Lewls are shown In cross section in figure 4. The difference in surface rough-

ness after sanding can be seen by comparing the as-sprayed coatings in fig-

ures 4(a) and (c) (specimens TNI and TKI) with the coatings after sanding shown

In figures 4(b) and (d) (specimens TN2 and TK2). The nonuniformlty In the bond

coat thickness Is apparent in all four specimens. It appears that large bondcoat powder particles were not fully melted during plasma spraying. There also

appears to be a significant amount of porosity in the ceramic top coat, espe-

cially in the two thicker coatings. Much of this porosity is believed due to

pull out during the pollshlng process even though the specimens were vacuum

epoxy impregnated prior to polishing. Some porosity is also apparent in the

bond coat which may also be due, in part, to the polishing procedure. APS bond

coats can contain signlflcant amounts of oxide resulting in poor adhesion of

small particles in the coatlng.

Cermet coatln_s. - Two cermet coatings applied by different techniqueswere tested. The metallic bond coat and cermet layers of one of the coatingswere applied by LPPS techniques while both layers for the second coating wereapplled by APS techniques. The volume fraction of ZrO2-Y203 in the APS cermetwas approximately 0.33 whereas the volume fraction in the LPPS cermet wasapproximately 0.47. The tubes for the cermet coatings were _nstrumented with asingle thermocouple attached to the inner wall at the midpoint (lengthwise) ofthe tube. Coating designations, thicknesses, compositions, and surface rough-nesses are given In table III.

The post-test mlcrostructure of the LPPS 50/50CM coating is shown in fig-ure 5(a). The LPPS bond coat is almost indistinguishable from the Mar-M246+Hfsubstrate. It is also apparent from the mlcrostructure that some of the largeNi-Cr-AI-Y particles in the outer cermet coating were not fully melted duringplasma spraylng. Some porosity is also apparent in the ZrO2-Y203 particles inthe cermet layer. Because this porosity is on a much finer scale than thatfor the traditional TBC's shown _n figure 4, It is not apparent whether the

2The nominal composition of B-1900 superalloy is Ni-IOCo-SCr-6Mo-6AI-4Ta-]Ti-O.lC-O.IZr-O.Ol5B wt %.

porosity is due, In part, to polishlng or was present followlng plasma spray-

ing. The post-test microstructure of the APS 30/70CM coating is shown In fig-

ure 5(b). Oxide stringers, generally parallel to the surface, are apparent in

the bond coat and in the cermet coating. These oxide stringers generally

result from the metal powder particles developlng a thin oxide scale duringplasma spraying in air. The molten powder particles Impact the surface and

"splat" developlng the traditional "pancake" or "splat" morphology with theoxide stringers dellneating the splat boundaries.

Measured Temperature Profiles

The effectiveness of the traditional TBC's applled at NASA Lewis Is

apparent In figure 6(a) where the temperature profile for an uncoated tube iscompared to representative temperature profiles measured below TBC's of two

different thicknesses. The addition of lO0 to 125 Nm ceramic layer for the TNI

specimen reduced the maximum metal temperature by approximately 350 °C. Sur-

prlsingly, increasing the ceramic layer thlckness to 200 to 225 Nm for the TKI

specimen only resulted in an additional metal temperature reduction of I00 °C.The rapid temperature decrease of the uncoated tube between 6.3 and 6.6 sec

occurs during purging with H2 and the more gradual cooling after 6.6 sec occurs

during purging with N2. Even with the TBC's, the effectiveness of the H2 purgeis apparent.

Representative temperature profiles for the cermet coatings and that forthe tradltlonal TBC (specimen TK3) with thermocouples on the inner wall of the

tube are shown in figure 6(b). The near-surface temperature profile for the

uncoated tube, shown for comparison, Indicates that the cermet coatings offeronly llm|ted thermal protection.

THERMAL MODELLING ANALYSIS AND RESULTS

Thermal Models for Coated Throat Tubes in the NASA Lewis Rocket Engine

Thermal model development and analysls. - Two finlte-difference thermalmodels were developed to simulate heat transport in coated and uncoated tubesand rods during testlng in the rocket engine. The first thermal model simu-lated convective heat transfer at the stagnation point and accounted for con-ductive heat transfer in a radlal direction only (one-dimenslonal model). Thesecond thermal model considered convective heat transfer over the surface ofthe tube from the stagnation point (e : 0°) to the exhaust side of the tube(e - 180 ° ) and accounted for both radlal and tangentlal conductive heat trans-fer within the tube wall (two-dlmenslonal model). The two models are shownschematically in figures 7(a) and (b).

The one-dlmenslonal thermal model consisted of a 45 ° sllce which contained

61 nodes for the tube and 86 nodes for the rod. For the tube, the near-surface

grid spacing was 12.5 pm (0.0005 In.) and the spacing near the inner wall was

50 pm (0.002 in.). The near-surface grid spacing for the rod was the same asthat for the tube increasing to 125 Nm (0.005 in.) at the center of the rod.

The two-dlmenslonal thermal model consists of a total of 110 nodes and equidis-

tant grld spacing as shown in figure 7(b). The appropriate heat transfer equa-

tions for conduction, convectlon and radiation were solved for both models by

the numerlcal computer program known as SINDA (Systems Improved Numerlcal

5

Differencing Analyzer) (ref. 4). The SlNDA program allows considerable flexl-blllty, including temperature-dependent thermal conductivitles (K(T)), specificheats (Cp(T)), and emlsslvlty values (_(T)), as well as time-dependent gas tem-peratures and heat transfer coefficients (hc). The temperature dependence ofK and Cp for the Mar-M246+Hf substrate (ref. 5) is shown In figures 8(a)and (b), respectively, and the density (p) of the substrate material (ref. 5)is given in table IV.

The gas temperature used in the one-dlmenslonal model was that for e : 0°(stagnation point), shown in figure 2. A value of 33 kW/m 2 °C was used for

the heat transfer coefficient at the stagnation point (ref. 3). During engine

start-up, the heat transfer coefficient was scaled with the normalized chamber

pressure as (Pc/Po)0.8 (Po = 2.07 MPa) to reflect the pressure dependence ofthe heat transfer coefficient assuming turbulent heat transfer (ref. 3). The

resultlng tlme dependence of hC durlng the 1.2 sec heatlng, H2 purge, and N2purge Is shown in figure 9(a). These values for hc were used very success-fully to predict near-surface metal temperatures In an uncoated Mar-M246+Hf

tube during testing In the rocket engine (ref. 3).

The convergence of the numerical solution was Inltially verifled with the

one-dlmenslonal model by varying the number of nodes and thereby changing the

grid spacing. Temperature proflles through the wall of an uncoated tube were

predicted with 16 and 30 nodes and compared to the predictions using 61 nodes.

The grid spacing near the surface with both 16 and 30 nodes was 25 Nm whereas

the spacing was 125 _m at the inner wall with 30 nodes and 500 Nm with

16 nodes. Temperatures throughout the tube wall dlffered by less than I °C

when the number of nodes was reduced from 61 to 30. However, when the number

of nodes was reduced to 16, surface temperatures were up 9° higher whlle inner

wall temperatures were up to 41 °C lower. Hence, the grid spacing resulting

from using 30 or more nodes was sufficient to produce a convergent solution.

The posslbility of tangential heat flow reducing the stagnation point tem-peratures, where metal temperatures were measured, was examined with the two-

dimensional thermal model. Figure 2 shows that the measured gas temperatures

on the exhaust side of the tube (e = 180 °) were slgnlfIcantly lower than thoseat the stagnation point. In addition, results from reference 3 indicate that

the heat transfer coefficient decreases in value from the stagnation point to

the exhaust side of the tube, as shown In figure 9(b). This combination of

high gas temperatures and hlgh heat transfer coefficients at the stagnation

point, and conversely, low gas temperatures and low heat transfer coefficients

on the exhaust side of the tube would be expected to produce much higher metal

temperatures at the stagnation polnt than on the exhaust slde of the tube.

This difference In metal temperatures could result in significant tangential

heat transfer within the tube wall reducing the metal temperature at the stag-

nation point where temperature measurements on coated tubes were made. How-

ever, measured gas temperatures and heat transfer coefficients at e= ±90 ° are

slmilar to those at the stagnatlon polnt (ref. 3) which should reduce any

Impact at the stagnation point of tangential heat flow to the exhaust side of

the tube. AddltlonaIly, the heat transfer coefficients decrease slgnlficantly

at approxlmately e : ±130 ° where flow separation occurred (ref. 3), such that

any large temperature gradients in the tangential dlrectlon would be expected

near this region.

The maximum effect of the tangential heat flow was evaluated with the two-dimensional model by conslderlng an uncoated tube and using gas temperatures

for e - ±0° (fig. 2) at nodes between angles 0° and 150 ° and the lower gas

temperatures For e = 180° (Fig. 2) for nodes between 150 ° and 180 °. Values

for hc at the lO surface nodes around the tube were taken from figure 9(b).

The two-dlmenslonal thermal model was run with and without tangentlal

heat flow. Predicted surface temperatures around the tube after 0.3, 0.6 and

1.2 sec, accounting for tangentlal heat flow, are shown in figure lO(a). As

expected, the surface temperature at the stagnation point (e - 0°) rises more

rapidly than that on the exhaust side of the tube (e = 180°), rising to more

than 80 percent of the maximum temperature after only 0.3 sec. The predicted

time dependence of the surface temperature at e = 0° and at e : 180 °, with

and without tangentlal heat flow, are shown in figure lO(b). There is less

than 10 °C reduction In the stagnation point temperature (e = 0°) when tangen-

tlal heat transfer is taken Into account. However, on the exhaust side of the

tube (e : 180°), heat flow around the tube to the exhaust side increases the

temperature approximately 140 °C. Hence, tangential heat flow does not slgnif-

Icantly reduce the stagnation point temperature but does substantially increase

the temperature on the exhaust side of the tube (e : 180°). Therefore, in com-

parlng measured and predicted stagnation point temperatures for tubes tested

In the NASA Lewis rocket engine, the one-dlmenslonal thermal model (45 ° wedge

facing the Injector) was considered adequate for predIctlng temperatures within

the scatter In the measured temperature data.

The effect of the range in the reported values of the thermophysical

properties of ZrO2-Y203 ceramic coatings was examined with the one-dlmenslonal

model. Several reported values of the temperature dependence for the thermal

conductlvlty (K) and the specific heat (Cp) for APS ZrO2-Y203 are shown in

flgures ll(a) and (b). The effect of this range In the values for K was

examined by predlctlng temperature profiles for a 180 Nm (0.007 in.) thickceramic layer on a Mar-M246+Hf tube. Two cases wlth different values for

K and Cp for APS ZrO2-Y203 were considered. For Case I, values for Kreported in reference 6_ (curve 12, fig. If(a)) and values for Cp reported

in reference 7 (curve lO, fig. ll(b)) were used (values for Cp were not

reported In ref. 6). For Case II, values for K and Cp reported In refer-

ence B (curve 9, figs. If(a) and (b)) were used. These values for K were

chosen as a lower bound (Case I) and as an upper bound (Case If) for the

reported data (fig. If(a)). The values for Cp for both cases are nearly

identical (fig. ll(b)). Identical K and Cp data for the Mar-M246+Hf sub-

strate were used. The predicted temperatures for the surface of the ceramic

and at the ceramic-metal interface (tube surface) are shown in figure 12. Fig-

ure 12 shows that the range in the values of the thermal conductivity amounts

to less than a 50 °C dlfference in the predicted ceramic surface temperature.

However, the predlcted temperatures vary by 180 °C at the ceramic/metal Inter-

face. Hence, the range In the reported values for K for APS ZrO2-Y203

coatings result in slgnlflcant temperature differences at the ceramic/metal

Interface and moderate temperature differences on the surface of the ceramic.

3Reference 6 gives a regression equation fitting data from several sources

including data for ZrO2-Y203 TBC's applied by electron beam, physical vapordeposltlon (EB-PVD). However, as shown in this reference, little difference

is observed in the reported thermal conductlvity between the APS and EB-PVD

coatlngs.

Radiation from the ceramic coating reduces the temperature of the surface.Liebert (ref. 9) has reported correlation equations for the emissivity for sev-eral thicknesses of APSZrO2-Y203TBC's in the temperature range 27 to 2500 °C.For a 180 _m (0.007 in.) thick coating, the work by Liebert suggests a nearlydecreasing value of the emissivity from 0.96 at 27 °C to 0.46 at 1500 °C.Theseemlssivity values were used in the thermal model to evaluate the reduc-tion in surface temperature due to radiative heat losses. This effect amountedto less than a lO °C decrease at the surface of the ceramic coating and analmost negligible decrease at the ceramic/metal interface. Hence, it is notnecessary to account for radiative heat losses from the ceramic coating duringthe short duration testing used in this study.

Most TBC's are fabricated with a metallic bond coat layer between theceramic layer and the substrate. These bond coat layers are typically 75 to125 pm (0.003 to 0.005 In.) thick and may be plasma sprayed in alr CAPS)or atreduced oxygen pressures (LPPS). The thermal conductivity of various reportedbond coats, both APSand LPPS,are shownin figure 8(a). It is apparent thatLPPSbond coats exhibit a muchhigher thermal conductivity than APSbond coats,and generally exceed that for the Mar-M246substrate. The lower values for theAPSbond coats can generally be attributed to the amountof oxide present inthese coatings.

The effect of the presence of a bond coat on the temperatures in thesubstrate was evaluated by modelling a TBCwith a 180 pm (0.007 in.) ceramictop coat above a lO0 Nm(0.004 in.) metal|it bond coat. The effect of differ-ences in the reported values of the thermophysical properties of metallic bondcoats was examined by again considering two cases. In Case I, values for Kreported in reference lO for an APSNI-Cr coating (curve 2, fig. 8(a)) andvalues for Cp reported in reference ? for an APSNi-Cr-Al coating (curve 3,flg. 8(b)) were used. In Case II, values for K and Cp reported in refer-ence 8 for an APSNi-Cr-AI-Y coating (curve l, figs. 8(a) and (b)) were used.At hlgh temperatures, the values for K for Case I are double those forCase II (compare curves l and 2, flg. 8(a)). Figure 13 shows the predictednear-surface temperatures for the two cases as well as predicted temperatureswithout a bond coat present. For case I, wlth a thermal conductivity closerto that of the Mar-M246+Hfsubstrate, the temperature In the substrate isreduced by a maximum of only lO °C while for case II, with a lower reported

thermal conductivity, the temperature in the substrate is reduced by approxi-

mately twlce this amount. In the ceramic coating, temperatures increased but

by smaller amounts than the decreases In the substrate. Since values for K

for LPPS coatings are generally greater than that for Mar-M246+Hf (fig. 8(a)),

the presence of LPPS bond coats would be expected to have a negligible effect

on the temperatures In the substrate or the ceramic coating. Hence, accounting

for the presence of the bond coat layer between the substrate and the ceramic

top coat makes only a small difference in the predicted temperatures In either

the ceramic layer or the substrate.

The difference in the thermal response between a tube and rod was examined

by predicting temperature profiles for both substrate geometries with lO0 and

200 pm (0.004 and 0.008 in.) thick ceramic coatings. The gas temperature andheat transfer coefficient used in this comparison were those for B = 0° (see

figs. 2 and 9(a)). Thermophysical propertles and density for the Mar-M246+Hf

substrate are as given in figures 8(a) and (b) (curve 13) and table IV.Because of the range in the reported values for K for ZFO2-Y203 coatings,

median values were used (see flg. If(a)). Values for Cp were taken from

reference 7 (curve I0, fig. 11(b)) and a value of 4.8 gm/cm 3 was chosen as atyplcal density (table IV). The temperatures predicted on the surface of theceramic layer and at the ceramlc/metal Interface for both the tube and rodgeometry are shown in figure 14(a). As shown, there Is very llttle differencebetween the tube and rod in the surface temperature of the ceramic where, after1.2 sec, the rod was 34 °C lower than the tube for the thinner coating and16 °C lower than the tube for the thicker coating. More signlflcant differ-ences are apparent at the ceramlc/metal Interface. At this interface, the tem-perature for the rod is 160 °C lower than the tube for the 100 Nm (0.004 In.)coating and approxlmately 145 °C lower than the tube for the 200 pm (0.008 in.)coating. Figure 14(a) also shows that the temperature at the surface of thecoating and at the coating/metal Interface Increases slightly faster for thetube than that for the rod during the steady state gas temperature reglonbetween 5.5 and 6.2 sec (see also fig. 2). This result simply reflects thegreater thermal mass of the rod. Hence, for the same gas temperature, themetal surface temperature of the rod is approximately 150 °C lower than thatfor the tube.

It is also interesting to examine the temperature differences due to the

coating thickness as shown in figure 14(a). A temperature difference of 60 °C

exists at the ceramic surface of the two coatings after 0.3 sec, regardless of

the geometry. However, temperature differences for the two coating thicknesses

are greater at the ceramic/metal interface where, for the tube, this differ-ence due to the coating thickness Is 232 °C, and for the rod, the difference

Is 216 °C. For comparison, the predicted temperature of the surface of an

uncoated Mar-M246+Hf tube after 1.2 sec, under the same conditions, Is 1374 °C

(ref. 3). Hence, the predicted benefit of a lO0 Nm (0.004 in.) TBC In reduclngthe metal temperature for the tube substrate Is 294 °C and the benefit of a

200 Nm (0.008 in.) TBC Is 526 °C. As expected, the predicted temperature bene-fit is not directly proportional to the TBC thickness.

Radlal temperature profiles through a 200 Nm (0.008 in.) thick coating

and into rod and tube substrates are shown In figure 14(b). Simply by consld-erlng the area under the temperature profile for each substrate shows that the

rod acts as a greater heat slnk than the tube. It appears that the temperatureacross the ceramic coating decreases linearly. However, a close examlnatlon ofthe temperatures indicates a slight concave downward curvature consistent with

a nearly-constant surface temperature (see fig. 14(a)) and a slightly increas-

Ing thermal conductivity for the ceramic layer (see flg. If(a)). As the sur-

face temperature rises at short times (

(table IV). Becauseof the similarity In values (ref. 11), the thermal conduc-tlvity, specific heat, and density of Mar-M246+Hfwas used for the B-1900 sub-strafe (specimen TK3). The coating thicknesses of lO0, 140, 200, and 250 Nmwere chosen to span the measuredcoating thicknesses (table II). The measuredand predlcted temperatures 0.0381 cm (0.015 In) below the ceramic/metal inter-face for specimens TKI and TK2 (200 and 250 _m thick coatings) are showninfigure 15(a) and for the TNI and TN2(lO0 and 140 pm thick coatings) in fig-ure 15(b). The measuredand predicted Inner wall temperatures for the TK3specimen (250 Nmthick coating) are shownin figure 15(c). The cooling periodassociated with the H2 purge (6.2 to 6.5 sec) is predicted somewhatsooner thanactually observed. This difference in the onset of cooling may indicate thefinite tlme required for the 02 valve to close and for the H2 purge to actually

begin. The measured temperature profiles Indicate that the hot gas may be

present in the chamber for approximately 0.1 sec longer than the value used inmode111ng (I.e., an actual hot test duration of 1.3 sec). However, In consid-

erlng the range in the reported values for the thermal conductivity of the

ceramic layer, the agreement between the measured and predicted temperaturesIs quite good.

The temperature dependence of K and Cp for the cermet coatings Isgiven In figures 8(a) and (b), respectively. Average values for the coatingthicknesses of 64 pm (0.0025 in.) for the LPPS 50/50CM and 180 pm (0.007 in.)for the APS 30/70CM coatings (see table III) were used in the model. Measuredand predicted inner wall temperatures for these two coatings are shown In flg-ures ]6(a) and (b). Again, the agreement Is quite reasonable considerlng thenumber of potential sources of uncertainty.

Potentlal Benefit of TBC's on HPFTP Blades

Thermal model development. - A thermal model was developed to predict

heat transfer to the leading edge (LE) of an HPFTP blade during start-up ofthe SSME. The leading edge was selected because heat transfer coefficients In

this region are predicted to be greater than for other parts of the blade

(ref. 12). Thls LE thermal model approximated the leading edge with a cylin-der, the radius of the cylinder being equal to the radius of curvature of the

leading edge of the blade near the blade tip (radius : 0.0762 cm (0.03 In.)).

Only heat transfer to the stagnation point of the leading edge was consid-

ered. Similar to the one-dimenslonal model of the rod, the LE thermal model

used a 45° wedge from the cylinder to simulate heat transfer at the stagnatlon

point and consisted of a minimum of 20 nodes w_th a grid spacing of 0.0051 cm

(0.002 _n.) near the cylinder center, 0.0025 cm (O.OOl in.) near the metal

surface, and 0.00127 cm (0.0005 In.) in the coating. Coating thicknesses of

64 and 165 _m (0.0025 and 0.0065 in.) were considered. The LE model also utl-

IIzed the SINDA thermal differencing analyzer to solve the appropriate differ-

ential equations. The model is shown schematically In flgure 17.

The gas temperature profile for the HPFTP during englne start-up and theheat transfer coefficients predicted for the stagnation point of the leadingedge are shown in flgure 18 (refs. 12 and 13). The heat transfer coefficlentduring the first thermal excursion is low because of low pressures In the HPFTPpreburner and low rotational velocities for the turbine wheel. Values for K,Cp and p for Mar-M246+Hf were taken from figures 8(a) and (b) and table IV.

lO

Predicted temperatures. - The abillty of numerous coatings to reduce thethermal transients at the leading edge of an HPFTP blade was examined. The

range in the values for K for the traditional ZrO2-Y203 TBC's was examined

by again taking the data reported in reference 6 as a lower bound and the data

reported in reference 8 as an upper bound (see fig. 11(a)). Both cermet coat-

Ings were also examined as well as an APS Ni-Cr-AI-Y metallic coating usingdata for K and Cp reported in reference 8 (curve I, flgs. 8(a) and (b)).

The temperature dependent thermophysIcal properties (K and Cp) for each of

these coatings, as shown in figures 8 and 11, were used in the LE thermalmodel.

Predicted temperature profiles at the coatlng/metal interface for each of

the 165 pm (0.0065 in.) thick coatings are shown in flgure 19. The surface

temperature for an uncoated blade is shown for comparison. It is apparent that

significant temperature reductions (>lO0 °C) can be achleved with any of thecoatlngs. Temperature reductions are somewhat greater for the flrst thermal

excursion, however, the proportionate reduction for both thermal excursions is

similar. The maximum metal temperature at the first and second thermal excur-

sion is shown in figure 20(a) for the 165 pm (0.0065 in.) thick coatings and

figure 20(b) for the 64 pm (0.0025 in.) thick coatings.

DISCUSSION

The measured temperature profiles for the TNI and TKI speclmens (see

fig. 6(a)) indicate the significant potential of TBC's to reduce metal tempera-

tures during short thermal excursions. These temperature profiles indicate

that a temperature reduction of approximately 350 °C can be reallzed with only

a lO0 to 125 pm thick TBC layer. It is somewhat surprising that almost dou-

bling the thickness of the TBC to 200 to 225 pm only reduces the temperature by

an additional lO0 °C. The predicted temperature proflles for the two coating

thicknesses shown In figure 14(a) would Indicate that a larger temperature

reduction would be expected on doubling the coating thickness. Since tempera-

ture changes are governed by the nonsteady state heat equation and not a simple

linear or |nversely linear relatlonshlp, doubling the coatlng thlckness does

not double the thermal protection. Most likely, the smaller than expected tem-perature reduction can be attributed to the uncertainty In the coating thick-

nesses above the thermocouples (see table II).

The measured temperature profiles for the cermet coatlngs (see fig. 6(b))

indicate very little thermal protectlon is afforded by these coatings. How-

ever, the thermophysIcal properties of these coatings, specifically the thermal

expansion, should be intermediate between that of the substrate or metallic

bond coat and a ceramic coating layer. The difference in thermal expansionbetween a substrate and ceramic coating is believed to be a major cause of

spallatlon of the ceramic layer during thermal cycling of TBC's (ref. 14). The

cermet layer may, therefore, be a useful transition layer between the substrate

and ceramic top coat. As a practical consideration, the cermet coating must

remaln relatively thin so as not to add undue welght to the blade. The LPPS

cermet coatings show that relatively thin coatings can be applied with thick-ness uniformity, at least to the simple substrate geometries examined in this

study.

Thermal modelling allows the effect of varlous model conflguratlons (e.g.,

rod versus tube, wlth and without a bond coat) and uncertalntles in the values

II

of the thermophyslcal properties (e.g., K(T)) to be examined and quantified.For a I00 pm (0.004 in.) APS metallic bond coat, where the reported values forK dlffer by more than a factor of two, it was shown that the substrate temper-atures were reduced by only a maximum of I0 to 20 °C while temperatures in theceramic increased even less. This limited effect of the bond coat layer isexpected by considering the overall thermal resistance of the TBC which is sim-ply the sum of the thermal resistance for each layer (ceramic and bond coat).Since the thermal resistance Is inversely proportional to the thermal conduc-tivity, the thermal resistance of the ceramic (low K) outweighs that for thebond coat for similar layer thicknesses. The relatively high thermal conduc-tivity of LPPS bond coats, similar to that of the Mar-M246+Hf substrate,results in very little thermal resistance and would produce little effect ontemperatures In the substrate.

The uncertainty (approximately ±50 percent) in the reported values for

the thermal conductivity of the APS ZrO2-Y203 ceramic layer has a significant

effect on the predicted substrate temperatures and, to a lesser extent, the

temperature in the ceramic layer itself. It is not known whether the differ-ences in the values for K are solely due to variations in the ceramic mate-

rial (e.g., extent of porosity and microcracking, etc.) or to differences in

the measuring techniques. The small differences in the composition of the APS

ZrO2-Y203 ceramic cannot account for the differences in K. The data reportedin reference 6 indlcates only small differences between coatings applied by

air plasma spraying (APS) and coatings applied by electron beam, physical vapor

deposi'tlon (EB-PVD), even though the two techniques result in greatly differ-

Ing mlcrostructures and somewhat different compositions. Thermal conductivity

measurements of plasma sprayed ceramics are generally made by either laser

flash or steady state comparator techniques, both of which Involve difficul-

ties. Similarly, for bond coat materials, there is also a significant differ-

ence between the values for K and p reported in reference 8 and that

reported in references 7 and 15 which cannot be attributed to the composi-tlonal differences. For instance, the LPPS Ni-Cr-AI-Y bond coats reported in

reference 8 exhibit negligible poroslty (see the LPPS bond coat below the LPPS

50/50CM cermet coating in fig. 5(a)), yet the reported density of these coat-

ings is more than 25 percent less, and the values for K are more than 50 per-

cent less than that reported for other LPPS Ni-Cr-Al-base coatings (table IV).

Hence, additional thermophysical property measurements accompanied by better

coatlng characterization are required to resolve the somewhat large differ-

ences in the properties of seemingly similar coatings. However, the good

agreement between the measured and predicted temperatures for the NASA Lewis-

applied TBC's suggests that thermal modelling of these TBC's can be success-

fully performed using medlan values for K.

Modelling the thermal response of a coated rod and comparing it to that ofa coated tube allows the temperature difference due to the greater thermal mass

of the rod to be quantified. In past durability testing of TBC's in the rocket

engine, coated rods were tested at gas temperatures estimated to be 150 to

200 °C greater than that for the coated tubes. Even at this greater tempera-ture, the coating lives on the rods were generally greater than those for the

tubes. The results presented in figure I4(a) show that at the same gas temper-

ature, the temperature at the ceramlc/metal interface is 145 to 160 °C lower

for the rod than for the tube. Temperature profiles in a coated rod were also

predicted for higher gas temperatures. For a lO0 pm (0.004 in.) thick TBC,these predictions show that the gas temperature for the coated rod must be

250 °C greater than that for the coated tube in order to produce similar metal

12

surface temperatures for both substrate geometries. Likewise, for a 200 pm(0.008 in.) thick coating, the gas temperature for the coated rod must beincreased 310 °C over that for the coated tube to produce simllar metal sur-

face temperatures.

One possible explanation for the increased durability of TBC's on the rod

substrates durlng durability testlng (ref. 2) can be understood by considerlng

the stresses whlch develop between the TBC and the substrate. Spalling of the

ceramic layer is believed due to differences in thermal expansion between the

ceramic layer and the underlying metallic substrate. Generally, a stress free

temperature Is defined as the temperature of the substrate during application

of the ceramlc layer. Cooling a coated component below this stress free tem-

perature results in a compressive load on the ceramic and tensile load on the

metallic bond coat. Testing at temperatures above the stress free temperature

results In a tensile load on the ceramic layer while the bond coat, constrained

by the ceramic layer, Is compressively loaded. High temperatures at theceramlc/metal interface which occur during testing can allow the relatively

soft metallic bond coat to plastically deform by thermally-activated processes

to reduce the compressive stress. The consequence of any relaxation by the

bond coat Is that the stress free temperature increases. Higher bond coat tem-

peratures allow the bond coat to deform more easlly and result In a higher

stress free temperature. Upon cooling, this increase in the stress free tem-

perature causes a greater compressive stress on the ceramic layer which can

result in cracking and spall|ng of the layer. Hence, the higher ceramlc/metal

interface temperatures encountered by the tube in comparison to that by the rod

can explain the shorter TBC lives on the tube substrates.

There is also some evidence that the TBC's can spall during heating In a

high heat flux environment, even durlng the first thermal cycle (ref. 2). In

this case, the higher ceramlc/metal interface temperatures produce a greaterstress at this interface due to the thermal expansion mlsmatch. This Increased

stress may cause sufficient damage within the TBC to result In spalllng.

The good agreement between the predicted and measured temperatures foreach of the various TBC's verifies the ability of the thermal model to simulate

heat transfer to coated throat tubes during short thermal excurslons. Good

agreement was observed at locations near the ceramic/metal interface and at the

_nner wall of the tube. For the temperatures at the ceramic/metal interface

(figs. 15(a) and (b)), the predicted temperatures maximize O.l to 0.2 sec

before the measured temperatures. This difference is due to the instantaneous

temperature decrease in the gas temperature profile used in the thermal model

(see fig. 2.). Although the oxygen valve begins to close after 1.2 sec, thevalve does not close Instantaneously and a finite time is required for removal

of the hot gas from the combustion chamber. It appears that a slightly longer

hot fire duration along with a somewhat more gradual temperature decrease would

be closer to the actual hot gas temperature proflle. However, the good agree-

ment justifies the use of the measured thermophyslcal property data for eachof the coatings and the use of thermal models to predict heat transfer during

short thermal excursions.

The predicted temperature reductions for coated HPFTP blades (figs. 19and 20(a) and (b)) indicate the tremendous potential of TBC's to reduce metal

temperatures during the thermal excursions associated with the start-up of the

SSME. As expected, the temperature decreases are ordered according to the

13

thermal conductivity with the largest decrease associated wlth the lowest ther-mal conductivity. For the 165 Nm(0.0065 In.) thick coatings, the temperaturereductions for the first thermal excursion vary from approximately 150 °C forthe LPPS50/50CMcoating to 470 °C for ZrO2-Y203TBC's using the thermal con-ductlvity data reported in reference 6. The temperature reduction for the APSZrO2-Y203TBC's using the thermal conductivity data reported in reference 8 isapproximately 335 °C, 135 °C less than that using the data reported In refer-ence 6, even though the APS ZrO2-Y203 coatings are essentially identical incomposition. As previously discussed, this difference indicates the need for

additional thermal conductivity data for plasma sprayed coatings and a better

understanding of the parameters which affect the thermal conductivity of plasma

sprayed coatings. However, even though significant differences exist in the

reported values for the thermal conductivity and predicted temperatures, the

results shown in figures 20(a) and (b) indicate the significant potential ofTBC's to reduce the thermal shock to the blades in the HPFTP.

SUMMARY AND CONCLUSIONS

Traditional APS ZrO2Y203 TBC's and APS and LPPS cermet coatings were

tested In a H2/O 2 rocket engine located at NASA Lewis. The traditional

ZrO2-Y203 TBC's showed considerable metal temperature reductions durlng testingin the hydrogen-rlch environment. The thermal response during testing of each

of the coatings has been successfully modelled. Good agreement was observed

between predicted and measured temperatures at the inner wall of the tube and

in the metal near the coatlng/metal interface.

The thermal model was also used to examine the effect of the difference

in the reported values of the thermal conductivity of plasma sprayed ZrO2-Y203

ceramic coatlngs, the effect of a lO0 _m (0.004 in.) thick metallic bond coat,

the effect of tangential heat transfer around the tube, and the effect of radl-ation from the surface of the ceramlc coating. It was shown that for the short

duratlon testing in the NASA Lewis rocket englne, radiation from the ceramiccoating, tangential heat transfer within the tube wail, or the presence of a

lO0 pm (0.004 in.) bond coat do not significantly affect the predicted metal

temperatures. However, the difference in the reported values for the thermal

conductlvlty of traditional APS ZrO2-Y203 TBC's can amount to significantlydifferent temperatures (>lO0 °C) predicted in the tube. The thermal model was

also used to predict the thermal response of a coated rod and compared to that

for a coated tube. Results from this comparison yield the difference in the

metal temperatures for the two substrate geometries which can be used to

explain the increased life of coatings on rods in comparison to tubes.

A thermal model was also developed to predict heat transfer to the leading

edge of HPFTP blades during engine start-up. The ability of various TBC's to

reduce metal temperatures during two thermal excursions was predicted. Temper-ature reductions of 150 to 470 °C were predicted for 165 Nm (0.0065 in.) thick

coatings for the larger of the two thermal excursions.

Speciflc conclusions for this study are"

I. Tradltlonal APS ZrO2-Y203 TBC's significantly reduce metal temperatures

during short thermal excursions. However, APS and LPPS cermet coatings reduce

metal temperatures to a much lesser extent than tradltional TBC's under thesame test conditions.

14

2. Thermal models using median values of the thermal conductivity of APSZrO2-Y203 TBC's were used to predict successfully metal temperatures below NASALewls-applled APS ZrO2-Y203 TBC's durlng testing in the H2/O 2 rocket engine atNASA Lewls.

3. Thermal models based on flnlte-difference techniques were successfullyused to predict metal temperatures below ZrO2-Y203/N|-Cr-AI-Y cermet coatedtubes tested in the NASA Lewis rocket englne.

4. Based on the successful modelling efforts stated in conclusions 2 and 3above using avallable thermophyslcal data for various TBC's, further modelllngresults indicate s_gnificant temperature reductlons (>100 °C) can be ach|evedat the leading edge of HPFTP blades during the thermal excursions which occuron SSME start-up.

ACKNONLEDGMENTS

The author would like to thank R.R. Holmes of NASA Marshall Space FlightCenter, Huntsville, AL, and T.N. McKechnle of Rocketdyne Division of RockwellInternatlonal, Huntsville, AL, for supplying the LPPS ZrO2-Y203/NI-Cr-AI-Y cer-met coatings.

REFERENCES

I. Holmes, R.R.: Vacuum Plasma Coatings for Turbine Blades. Advanced HighPressure 02/H 2 Technology, S.F. Morea and S.T. Wu, eds., NASA CP-2372,1985, pp. 74-90.

2. Brlndley, W.J.; and Nesbltt, J.A: Durability of Thermal Barrier CoatingsIn a H_gh Heat Flux Environment. Advanced Earth-To-Orblt Propulslon Tech-nology 1988, R.J. Richmond and S.T. Wu, eds., NASA Marshall, 1988, Vo1. I,pp. 661-674.

3. Nesbitt, J.A.; and Brlndley, W.J: Heat Transfer to Throat Tubes In aSquare-Chambered Rocket Engine at NASA LeRC. NASA TM-I02336, 1989.

4. Fink, L.C.: SINDA-Systems Improved Numerlcal Differencing Analyzer.Available through Computer Software Management and Information Center(COSMIC), The University of Georgla, Athens, GA, 30602, Program MSC 18597.

5. Chandler, W.T.: Materials for Advanced Rocket Engine Turbopump Blades.(RI/RD83-207, Rockwell International Corp., NASA Contract: NAS3-23536)NASA CR-174729, 1983.

6. Strangman, T.E.; Neumann, J.; and Liu, A: Thermal Barrier CoatingLife-Predictlon Model Development. NASA CR-179648, 1987.

7. Brandt, R., et al.: Specific Heat and Thermal Conductivity of PlasmaSprayed Yttrla-Stabll|zed Z|rconla and NIAI, NiCr, NiCrAI, NiCrAIY,NICoCrAIY Coatings. High Temp.-High Pressures, vol. 18, no. I, 1986,pp. 65-77.

15

8. Holmes, R.R.; and McKechnle, T.N.: Vacuum Application of Thermal Barrier

Plasma Coatings. Advanced Earth-to-Orblt Propulsion Technology, 1988,

R.J. Richmond and S.T. Wu., eds., NASA CP-3012, 1988, Vo1. I, pp. 692-702.

9. Llebert, C.H.: Emittance and Absorptance of the National Aeronautics and

Space Adminlstratlon Ceramic Thermal Barrier Coatlng, Thln Solid Films,vol. 53, no. 2, 1978, pp. 235-240.

10. Wilkes, K.E.; and Lagedrost, J.F.: Thermophyslcal Properties of PlasmaSprayed Coatings. NASA CR-121144, 1973.

11. Aerospace Structural Metals Handbook, Battelle Laboratories, Columbus, OH,1988.

12. Abdul-Azlz; Tong, M.T.; and Kaufman, A.: Thermal Finite-Element Analysisof Space Shuttle Maln Engine Turbine Blade. NASA TM-IO0117, 1987.

13. Abdul-Azlz: Private Communication. (Thts data was summarlzed In the pub-]]catlon given in ref. 12.)

14. M11]er, R.A.: Current Status of Thermal Barrier Coatings - An Overview.

Surf. Coat. Technol., vo1. 30, no. l, 1987, pp. 1-11.

15. DeMasl, J.T.: Thermal Barrler Coating Life Prediction Mode] Development.NASA CR-17950B, 1986.

16

TABLE I. - PLASMA SPRAYING PARAMETERS FOR NASA LEWIS COATINGS

[Environment, O.IO-MN/m 2 (l-atm) air; arc gas, 23-1iter/min argon: plasma gun, EPI 120-kW

plasma generator (#167 anode, #27 cathode, 0.040 stainless steel powder port).]

Parameter Bond coat Ceramic topcoat

Ni-35.8Cr-6.3Al-O.96Y Z r(]2-6.SY203Composition, wt %Tube rotation, rpm

Plasma gun:Power, kW

Traverse speed, mm/sNumber of passes

Standoff distance (gun to specimen), cm (in.)Power feeder:

Rotation, rpmGas, liter/min

14

302

II.4 (4.5)

48IO

a6 to 9

I0.2 (4)

2.0Ar, 1.6

aDepending on desired thickness.

TABLE If. - THERMAL BARRIER COATINGS APPLIED AT NASA LEWIS

Coating Bond coat thickness b Ceramic coat thickness b Surface

designation a roughness(RA) c

I

Before sanding

TNI 25 to 50 I to 2 lO0 to 125 4 to 5

TN2 (d) (d) (d) (d)TKI 25 to 50 l to 2 200 to 225 7.9 to 8.8

TK2 (d) (d) (d) (d)TK3 e 50 to 75 2 to 3 225 to 275 8.8 to I0.8

After sanding

9.4 37210 3959.3 3659.6 378

TN2 25 to 75 l to 3 lO0 to 140 4 to 5.5 I 3.2 I 127

TK2 25 to 50 l to 2 200 to 250 7.9 to g.8 1 5.1 I 200

aTN refers to coatings with a thinner ceramic top coat, TK refers to coatingswith a thicker ceramic top coat.

bMeasured from cross sections after testing.CAverage of Four measurements.dunmeasured prior to sanding.eB-IgO0 superalloy substrate.

TABLE III. - CERMET THERMAL BARRIER COATINGS

Bond coat thickness, clam (in.xlO-31Cermet coat thickness, c _m(in.xlO -j)

Surface roughness (RA), _ (_in.)

Volume fraction of _xideComposition (wt %):

Metal (in bond coat and cermet)Ceramic (in cermet)

Coating desi

LPPS 50/5OCM a

5O (2)50 to 75 (2 to 3)

5.3 (210)0.47±0.12

Ni-16.SCr-5.5Al-O.55Y

ZrO2-8Y203

gnation

APS 30170CM-5-

40 to 60 (1.6 to 2.4)

150 to 200 (6 to 8)9.4 (372)0.33±0.I0

Ni-16.5Cr-5.5AI-O.55Y

ZrO2-8Y203

aLow-pressure, plasma-sprayed bond coat and cermet coat (approx. 50 percent ceramic,50 percent metal in cermet coat). Applied by NASA Marshall Space Flight Center,Huntsville, AL.

bAit-plasma-sprayed bond coat and cermet coat (approx. 30 percent ceramic, 70 percentmetal in cermet coat). Applied by Plasma Coating Corporation, Gardina, GA.

Measured from cross sections after testing.Taken from reference 8.

17

TABLE IV. - DESIGNATION AND DENSITY OF COATINGS

AND SUBSTRATE

Designation a

APS metallic coatings:l Ni-Cr-AI-Y

2 Ni-Cr

3 Ni-Cr-AI

LPPS metallic coatings:4 Ni-Cr-AI-Y

5 Ni-Cr-AI

6 Ni-Co-Cr-A1-Y

Cermet coatings:7 LPPS 50/50CM

8 APS 30/70CM

APS ceramic coatings:

g ZrO2-8Y203

10 Zr02-(7 to 8)Y203

11ZrO2-7Y203

|2 ZrO2-Y203 e

Substrate:13 DS Mar-M246+Hf

Density,p,

g/cm 3

4.48

6.8

6.9

5.48

7.48 b

4.55

4.01

3.47

5.39 c

4.5 d

8.6

Reference

8

I0

7

8

7

15

8

7

14

6

aNumber used to identify substrate and coatingin figures 8, 11, Ig, and 20.

bAverage for three different LPPS Ni-Cr-Al-based

metallic coatings.CRepresentative values for several reported

d APS Zr02-(7 to 8)Y203 coatings.Based on reported 20 percent porosity.

eRegression fit of data from several sources

includes data for ZrO2-Y203 coatings appliedby electron-beam, physical vapor deposition.

IIYDROGEHMANIFOLD_ INJECTOR TtlROAT I_12.1 cM _j

_iG_XIYTIGEoNNMANIFOL')-_\y-[_" ----7 ....... _r _[o_TIIoN_-- _-

( _ (- _ _ - SOUARECOMBUSTION \ ./

FIGUREI. - SCHEMATICVIEWOF TIIEROCKETENGINEAND COPPERSPECIMENIIOLDER.

CM

18

1600

lqOO

12oo

IOOO

P- 800

GO0

400

200

9, DEG

I/ -- 0 STAGNATIONPT.

Ir ....... 18oli fo-_

/A,,

5.0 5,5 G,O

TIME, SEC

._J..---L=--

6.5

FIGURE 2, - GAS TEMPERATUREPROFILES AT THE STAGNATIONPOINT

(0 = 0°) AND AT THE EXHAUSTSIDE (0 _ 180°) OF TEST POS[-

TION 2 NEAR THE TUBE MIDPOINT (LENGTIIWISE) (REF, 3).

||i

I[]II

i

200 pm

FIGURE 3. - CROSS-SECTIONAL VIEW OF TIIEEAR-SURFACE TIIERMOCOUPLESBELOW A TRADITIONAl

ZFO2-Y203 TBC.

O_IGINAE PAGE

BLACK AND WHILE PH.OTOGRAPHI_

19

ORIGINAL PAGE

BLACK AND WHITE PHOTOGRAPH

[a) TN1.

(b) TN2.

FIGIJRE It. - CROSS-SECTIONALMICROSTRIJCTIJRESOF THE NASA-LERC APPLIED ZrO2-Y203 TDC'S.

2O

ORrGINAL PA(3E

BLACK AND WHITE PHOTOGRAPH

(c) TK1.

!tOO IJm

I

(d) TK2,

FIGURE q. - CONCLUDED.

21

ORIGINAl. PAGE

BLACK AND WHITE PHOTOGRAPH

(a) LPPS 50/50CM,

(b) APS 30/70CM.

FIGURE 5, - CROSS-SECTIONAL MICROSTRUCIURES OF THE CERMET TBC'S,

22

i.-,

1400

1200

1000

BOO

600

400

20O

i

I

i

! I L......]_

,,-UNCOATED/

I I I

35

>1I--

=_I,--

30

25

20

15

lO

5

........ LPPS METALLIC ,/' G.... APS METALLIC

/7'

..... LPPS CERMET/i

.......... APS CERMET ,,'DS MAR-M245+Hf ,/" - 5

7"

..- I3

.._ - - .-. ..................... _ - - - i

.i- _-L-_-L"--"---.... :..-..::.--...-....T.2..L-_.'L..'L................8:.:"__.-:...T:._.,'7............................................_ L__. J._L_ ...... !..... J __-L--J--0

(a) THERMAL CONDUCTIVITY (K) FOR THE APS AND LPPS METALLIC

COATINGS, AND THE DS MAR-M2hG+Hf SUBSTRATE.

700

o_

-_ 600

500u-

4O0

I ..-+-"G

8 ,,,,.'" ._.._;_;,x_-5 AND 5_ -" --_#_

/'_..--:." cS_.c_:;."I _ 13

0 200 400 600 800 1000 1200

TEMPERATURE.°C

(b) SPECIFIC HEAT (Cp) FOR THE APS AND LPPS METALLIC COA]-

INGS, THE CERMET COATINGS, AND THE DS MAR-M2h6+Hf SUB

STRATE.

FIGURE 8. - TEMPERATURE DEPENDENCE. COATING DESIGNATIONS

AND REFERENCES ARE GIVEN IN TABLE IV.

20

10

0

%

L, Q9.5 --

¢

-iI 1 4 1 I L I

S.O 5.2 6.2 6.4 G.G G.8

TIME, SEC

(a) TIME DEPENDENCE AT STAGNATION POINT (0 = 0°).

I I I I45 90 135 180

ANGLE FROM STAGNATION POINT, O

(b) ANGULAR DEPENDENCE AROUND 0.9525 CM (0.375 "IN.)TUBE

(REF. 3).

FIGURE 9. - HEAT TRANSFER COEFFICIENT.

24

1500

1250

1000

750

500

250

=_ 14oo--

1200--

1000 --

800 --

600 --

400 --

200 --

t = 1.2 SECONDS

__ t : O,G SECONDS

! = 0.3 SECONDS

f ^ 8oo 1oo0 _ x,

0o 1800

1 1 L...,_L_ 1 1 I I l20 40 60 80 100 120 140 160 180

ANGLEFROM STAGNATION, 0

(a) ANGULARTEMPERATUREDISTRIBUTION ON THE SURFACEOF THE

TUBE AT VARIOUS TIMES WITH TANGENTIAL HEAT FLOW,

0

tl / =

_-0 = 1800

TANGENTIAL HEAT FLOW

NO TANGENTIAL HEAT FLOW

o I I4.8 5,8 G.8

TIME, SEC

(b) TIME DEPENDENCE OF THE SURFACE TEMPERATURE AT 0 = 0 (STAG-

NATION POINT) AND AT 0 - 18G°, (EXHAUST SIDE OF THE TUBE),

WITH AND WITHOUT TANGENTIAL _IAT FLOW,

FIGURE I0, - PREDICTED TEMPERATURE PROFILES FOR AN UNCOATED

TUBE USING THE 2-D MODEL.

o_

E

E

W:

¢_}

i.5

1.0

.5

o

700

G00

50O

4O0

MEDIAN K(T)

11.. -.............. --_._

o_

1600 --

1400 --

1200 --

1000 --

800 --

600 --

400 --

200 --

04.8

1000

800

600

0q.8

I

------ CASE II i""_

A/,//_ lqOC,

// _ "I 1200

s18. 618. 6ooTIME, SEC

nooFIGURE 12. - PREDICTED TEMPERATURE PROFILES SHOWING THE

EFFECT OF THE RANGE IN THE VALUES OF THE THERMAL CON-

DUCTIVITY FOR THE ZrO2 Y203 CERAMIC COATING (SEE TEXT

FOR A DESCRIPTION OF THE PARAMETERS USED IN EACH CASE).

NO BOND COAT r---_

CASE l

CASE II

/ -,I ,,07-

/ T'"J,s c!,5.8 (;.8

TIME, SEC

FIGURE 13. - PREDICTED TEMPERATURE PROFILES SHOWING THE

EFFECT OF A 100 I/m (0.004 IN.) METALLIC BOND COAT BELOW

THE zro2-Y205 CERAMIC TOP COAT. PREDICTED TEMPERATURESARE FOR A LOCATION 0.038 CM (0.015 IN.) BELOW THE SUB-

STRATE SURFACE OR BELOW THE BOND COAT/SUBSTRATE INTER-

FACE (SEE TEXT FOR A DESCRIPTION OF THE PARAMETERS USED

IN EACH CASE).

200

o_

_3

TUBE

__ ROD

CERAMIC

o I I4.8 5.8 6.8

TIME, SEC

(a) TEMPERATURE VERSUS TIME FOR THE CERAMIC SURFACE AND AT

CERAMIC/METAL INTERFACE,

1400

1200

1000

8O0

600

qO0

200

1400

,- , 1200

_ I

W I_-----__ 1ooo

-:_'-i-------_--1.NE,800"-.I TUBE

"""-, WALL GO0

-- TBC I FV_R-M246+Hf

-- N\

J C--r---J.01 .02 .03 .04

"--- RODTUBE ""-

_ CENTER-- ROD ....

I I I I I.1 .2 .3 .4 .5

DISTANCE, CM

(b) RADIAL TEMPERATURE PROFILES FORT HE 200 Nm (0.008 _N.)

THICK TBC's.

FIGURE 14. - PREDICTED TEMPERATURE PROFILES FOR A 100 AND

200 _m (0.004 AND 0.008 IN.) ZrO2-Y203 TBC ON A TUBE ANDROD SUBSTRATE.

26

1000 --

8OO

600 -- "_400

/' ..... PREDICTED200 g,,

#

I I I(a) NEAR-SURFACE TEMPERATURES FOR TKI AND TK2.

o v I I l(b) NEAR-SURFACETEMPERATURESFOR TNI AND TN2.

800 --

600

400

O "" I

6 7 8

TIME, SEe

(C) INNER WALL TEMPERATURES FOR TK3.

FIGURE 15. - MEASURED AND PREDICTED TEMPERATURES FOR THE

LERC-APPLIED ZtO2-Y2% TBC's.

o

12oo

1ooo

800

600

400

200

200

!aO0 n

1200

1000

800

600

400

_' 200

_. o_ 1200

i1000 !

800

600

400

2O0

os_

m ij Sr_

///

/

_ ..... PREDICTED

a

" 1 I I I 1 I(a) LPPS 50/50CM

js-"_

js _

s S

-- //

o •

a_

m

/

' r i I i i 16 7 8

TIME, SEC

(b) APS 30/70CM

FIGURE 16. MEASURED AND PREDICTED INNER WALL TEMPERATURES

FOR THE CERMET COATED TUBES.

27

HALF CIRCLE AT /___ A/__,_LEADING EDGE

_ (RADIUS = 0'0762 CM)7

...,._i 4sDEGREESLICE20 NODESFOR UNCOATED

BLADE CROSSSECTION 23 NODESFOR 64 pmCOATING

NEAR TIP27 NODESFOR 165 pm

COATING

FIGURE 17. - SCHEMATICVIEW OF THE THERMALMODELFOR THE

LEADING EDGE OF THE HPFTP BLADE.

200 m 2000

E

150

125

100

75

__ so_ 25

175 --

1500

• I000

p_.

soo

O--

GAS

/

-- /

{/

FHERMALI / I', \ 1

/ ',.0 .---"" ,/ _''"J SECOND TIIERMAL

....,,--*"\wJ EXCURSION

soo I I I0 .5 1.0 1.5

TIME, SEC

FIGURE 18. - TIME DEPENDENCE OF THE GAS TEMPERATURE IN THE

HPFTP AND STAGNATION POINT HEAT TRANSFER COEFFICIENT FOR

THE LEADING EDGE OF AN HPFTP BLADE DURING SSME START-UP.

I2.0

1800

1600

1400

1200

1000

800

600 --

400

200 --

0 --

-2000

i,.-,._- I

-- I 1 GASI ,/'-

-- I Vi I

J _

1 -UNCOATED (13)

. /- 7 /- GAS

_- /_, ,--_-_I _"_-'//-DNCOATED_"_-_ (I,,V /'/_/ *-9

!ft/;/

I I I I.5 1.0 1.5 2.0

TIME, SEC

FIGURE 19. - PREDICTED METAL SURFACE TEMPERATURES DURING

SSME START-UP FOR THE LEADING EDGE OF AN HPFTP BLADE

WITH AND WITHOUT VARIOUS 165 pm (0.0065 IN.) THICK THER-

MAL BARRIER COATINGS.

28

1200

lOOO

800

60O

400

2OO

0

1200

lOO0

800

600

400

200

THERMAL

EXCURSION EMPERATURE

F REDUCTION DUE

FIRST

_ SECOND TO COATING_/f

/I

Ii!Il

._J

(a) WITH AND WITHOUT 165 _m (0.0065 IN.) THICK TDC's.

TEMPERATURE

REDUCTION DUE

TO COATING_

// l

,j- qr-IiIIf

UNCOATED LPPS APS APS ZrO2 7ro2-

MM2qG+IIf50/SOCM NICrAIY 30/70CM Y203 Y203

(13) (7) (1) (8) (9) (12)

COATING TYPES

(b) WITH AND WITHOUT 6q #m (0.0025 _N.) THICK TBC's.

FIGURE 20, - MAXIMUM LEADING EDGE METAL TEMPERATURES FOR

THE TWO THERMAL EXCURSIONS WHICH OCCUR IHIRINGSSME START-UP.

29

Report Documentation PageNational Aeronautics andSpace Administration

1. Report No. 2. Government Accession No,

NASA TM-102418

4. Title and Subtitle

Thermal Modelling of Various Thermal Barrier Coatings in a High

Heat Flux Rocket Engine

7. Author(s)

James A. Nesbitt

9. Performing Organization Name and Address

National Aeronautics and Space AdministrationLewis Research Center

Cleveland, Ohio 44135-3191

12. Sponsoring Agency Name and Address

National Aeronautics and Space Administration

Washington, D.C. 20546-0001

3. Recipient's Catalog No.

5. Report Date

December 1989

6. Performing Organization Code

8. Performing Organization Report No.

E-5184

10. Work Unit No.

505-63- l A

1 I. Contract or Grant No.

13. Type of Report and Period Covered

Technical Memorandum

14. Sponsoring Agency Code

15. Supplementary Notes

16. Abstract

Traditional APS ZrO2-Y203 TBC's and APS and LPPS ZrO2-Y203/Ni-Cr-AI-Y cermet coatings were tested in a

1-12/O2 rocket engine located at the NASA Lewis Research Center. The traditional ZrO2-Y203 thermal barrier

coatings (TBC's) showed considerable metal temperature reductions during testing in the hydrogen-rich environ-

ment. A thermal model was developed to predict the thermal response of the tubes with the various coatings.

Good agreement was observed between predicted temperatures and measured temperatures at the inner wall of thetube and in the metal near the coating/metal interface. The thermal model was also used to examine the effect of

the differences in the reported values of the thermal conductivity of plasma sprayed ZrO2-Y203 ceramic coatings,the effect of a 100 #m (0.004 in.) thick metallic bond coat, the effect of tangential heat transfer around the tube,

and the effect of radiation from the surface of the ceramic coating. It was shown that for the short duration

testing in the rocket engine, the most important of these considerations was the effect of the uncertainty in thethermal conductivity of traditional APS ZrO2-Y203 TBC's which can amount to significantly different tempera-

tures (> 100 °C) predicted in the tube. The thermal model was also used to predict the thermal response of a

coated rod in order to quantify the difference in the metal temperatures between the two substrate geometries andto explain the previously-observed increased life of coatings on rods over that on tubes. A thermal model was

also developed to predict heat transfer to the leading edge of HPFTP blades during start-up of the space shuttle

main engines. The ability of various TBC's to reduce metal temperatures during the two thermal excursions

occurring on start-up was predicted. Temperature reductions of 150 to 470 °C were predicted for 165 #m

(0.0065 in.) coatings for the greater of the two thermal excursions.

17. Key Words (Suggested by Author(s)) 18. Distribution Statement

Thermal modelling; Thermal barrier coatings; High heat Unclassified-Unlimited

flux; Rocket engine coatings; Cermet coatings; Subject Category 26

TBC's; Rocket engine testing

19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No of pages

Unclassified Unclassified 30

NASAFORM1626OCT86 *For sate by the National Technical Information Service, Springfield, Virginia 22161

22. Price*

A03